agentleFS
GitHubKnown publishergithub/replica/starters/shared/.agents/context-templates/SKILL.md

context-templates

Skill117 starsChanged 10 days agoMITView on GitHub

Shared context template library for setup-skills. Read when discovering context documents or their ownership. This library supplies research inputs, not active product context or a separate authoring workflow.

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep all seven root identities.

    line 48
  2. Distinctive:

    Do not physically split every template or add new owners, intake or authors.

    line 61

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Shared context templates

This directory is the only editable template library. It is copied with the starter beside .agents/skills/. Do not duplicate templates inside individual skills or infer approved context from templates themselves.

Documents and research sources

Plain setup selects every document below once. For a named consumer, read its required-context declaration, including any linked reference that owns that declaration, and select its required and conditional documents once. Do not maintain a second consumer dependency map here. Users can also name documents directly. This entrypoint is not itself a context document.

DocumentSource groups
DESIGN-SYSTEM.mddesign-system
PRODUCT-DESIGN.mdproduct
REPOSITORY.mdrepository
SEED-DATA.mdproduct, repository
RESPONSIVE-DESIGN.mdproduct
ACCESSIBILITY.mdproduct, repository
BROWSER-VALIDATION.mdrepository

Source groups identify research ownership. design-system and product select the six shared intake questions in setup-skills' references/source-workflow.md. The repository group selects automatic, bounded, read-only research of the current consumer checkout under that workflow, not a source/permission question. Repository-only selection needs no intake answers. Missing repository answers or file registrations do not block this local scope. Other groups' research is not automatically authorized by a file being local. Collect each applicable question once across selected documents. Preserve all restrictions and explicit skips, including legacy repository opt-outs, and honor content exclusions. Volunteered docs add leads, not a new required question. Bounded proposals live in PRODUCT-DESIGN.md and apply only where permitted, never as researched product policy. An empty source registry does not disable automatic local research. Excluded, missing or inaccessible evidence must produce specific limitations, not fictional facts or replacement research against a skip.

Optional owner topics

Keep all seven root identities. Authors may keep a document whole or move task-specific detail into directly linked owner topics when supported tasks can skip genuinely irrelevant detail. Choose boundaries by task applicability and cohesion: keep a decision's requirements, conditions, exceptions, examples and citations together. Keep short, cohesive guidance whole when splitting would only add navigation. Check evidenced consumer constraints, such as incoming anchors or a parser that requires a particular structure, before moving content; do not assume one consumer's constraints apply to others. Observed oversized or truncated reads can motivate review, but file size, consistency with other owners or a fixed topic count does not justify a split. Splitting alone does not establish faster discovery or better task outcomes. Preserve correct existing text and its authority and historical qualifications; relocation is not permission to rewrite policy or claim fresh verification. Do not physically split every template or add new owners, intake or authors. Roots retain authority, common rules, universal obligations, known gaps and direct links to every topic with concrete applicability conditions. New topics use the owner's exclusive topics/<DOCUMENT-STEM>/ directory within active context; existing linked paths need preservation and explicit inventory.

Each trigger covers normal successful work and readiness as well as relevant failure/recovery work. For example, supported component selection, opening and checking an ordinary preview, and adding valid fixtures must select their applicable details before a failure occurs. Avoid read-all lists and topic chains. Each topic keeps complete usable rules, examples, conditions, exceptions and citations. Shared-owner links transfer ownership, not duplicate policy.

Apply setup's complete-set authoring, preservation and approval contract to roots and topics together. Library templates are research prompts, not approved consumer guidance or permission to activate extra files.

Location

Setup reads this library from .agents/context-templates/ in the current prototype repository. It does not fetch or install another copy. Context documents become active only after separate explicit approval.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.